llm_context_core

Tip

WASM-compatible — compiles to both wasm32-wasip1 and native targets. Zero runtime dependencies beyond serde and log.

Crate: llm_context_core · Path: crates/llm_context_core
Description (Cargo.toml): WASM-compatible LLM context window management — token estimation, budget tracking, history strategies, and memory traits

Role

WASM-compatible LLM context window management library. Provides token estimation, context budget computation, pluggable history management strategies, and trait ports for long-term memory backends. Three-tier architecture:

• Tier 1 (Working) — current turn messages, bounded by token budget
• Tier 2 (Short-Term) — conversation history with sliding window / summarization
• Tier 3 (Long-Term) — cross-conversation semantic retrieval (Qdrant, etc.)

Feature flags

From crates/llm_context_core/Cargo.toml:

Feature	Purpose
default	Empty — no features enabled by default

Public API (from src/lib.rs)

Modules: budget, history, memory, strategy, tokens

Re-exports:

• ContextBudget — token budget computation
• HistoryManager, HistoryManagerConfig — conversation history management
• LongTermMemory, MemoryEntry, MemoryFilters, MemoryType — trait ports and types for long-term memory backends
• ContextStrategy, ContextStrategyKind — pluggable context strategies
• estimate_tokens — token count estimation

Example sketch

use llm_context_core::{ContextBudget, HistoryManager, HistoryManagerConfig, estimate_tokens};

let budget = ContextBudget::new(128_000);
let config = HistoryManagerConfig::default();
let mut history = HistoryManager::new(config);

let tokens = estimate_tokens("Hello, how are you?");

Full API reference: cargo doc -p llm_context_core --no-deps